completed unit tests

woltka/table.py

@@ -456,7 +456,7 @@ def table_add_metacol(table, dic, name, missing=''):
     """
     # redirect to BIOM module
     if isinstance(table, Table):
-        biom_add_metacol(table, dic, name, missing='')
+        biom_add_metacol(table, dic, name, missing=missing)
         return
 
     # add metadata column

woltka/tests/data/README.md

@@ -59,6 +59,15 @@ woltka classify \
   --output bowtie2.free.tsv
 ```
 
+`bowtie2.free.1p.tsv`:
+
+```bash
+woltka tools filter \
+  --input output/bowtie2.free.tsv \
+  --min-percent 1 \
+  --output bowtie2.free.1p.tsv
+```
+
 `blastn.species.tsv`:
 
 ```bash
@@ -146,6 +155,15 @@ woltka classify \
   --output split.process.tsv
 ```
 
+`merged.process.tsv`:
+
+```bash
+woltka tools merge \
+  --input output/burst.process.tsv \
+  --input output/split.process.tsv \
+  --output merged.process.tsv
+```
+
 `diamond.free.tsv`
 
 ```bash
@@ -169,3 +187,45 @@ woltka classify \
   --rank function \
   --output diamond.func.tsv
 ```
+
+`truth.gene.tsv`
+
+```bash
+woltka classify \
+  --input align/truth/b6o \
+  --coords function/nucl/coords.txt.xz \
+  --output truth.gene.tsv
+```
+
+`truth.uniref.tsv`
+
+```bash
+woltka classify \
+  --input align/truth/b6o \
+  --coords function/nucl/coords.txt.xz \
+  --map function/nucl/uniref.map.xz \
+  --names function/uniref.names.xz \
+  --map-as-rank \
+  --rank uniref \
+  --output truth.uniref.tsv
+```
+
+Or:
+
+```bash
+woltka tools collapse \
+  --input output/truth.gene.tsv \
+  --map function/nucl/uniref.map.xz \
+  --names function/uniref.names.xz \
+  --output truth.uniref.tsv
+```
+
+`truth.goslim.tsv`
+
+```bash
+woltka tools collapse \
+  --input output/truth.uniref.tsv \
+  --map function/go/goslim.tsv.xz \
+  --names function/go/name.txt.xz \
+  --output truth.goslim.tsv
+```

woltka/tests/data/output/bowtie2.free.1p.tsv

@@ -0,0 +1,30 @@
+#FeatureID	S01	S02	S03	S04	S05
+1028307	0	0	0	1888	0
+1125630	0	778	0	0	0
+1133852	0	0	864	0	0
+176299	0	0	22	0	0
+190304	0	26	0	0	226
+190485	0	0	0	0	88
+198214	0	0	42	0	0
+208435	0	0	0	22	0
+211586	0	672	0	0	0
+220341	0	0	0	0	436
+224324	0	0	50	0	0
+243160	132	0	36	0	0
+243231	0	0	0	0	120
+246196	176	0	0	0	0
+272560	48	0	0	0	0
+272621	0	0	22	0	0
+272631	0	0	0	0	86
+272947	0	0	54	0	0
+295405	0	270	0	0	0
+300852	1614	0	0	0	0
+333849	0	0	0	54	0
+365659	0	0	156	0	0
+366394	0	88	0	0	0
+380703	0	0	0	0	954
+393305	0	0	640	0	0
+413999	0	80	0	0	0
+441771	0	62	0	0	0
+525903	0	0	0	0	56
+585056	0	0	22	0	0

woltka/tests/data/output/merged.process.tsv

@@ -0,0 +1,205 @@
+#FeatureID	S01	S02	S03	S04	S05
+GO:0000160	0	6	2	0	0
+GO:0000162	2	0	0	4	0
+GO:0000724	0	4	0	0	0
+GO:0000746	0	0	3	0	0
+GO:0000917	0	0	0	0	2
+GO:0002949	0	0	1	0	0
+GO:0005975	16	4	6	4	6
+GO:0005978	4	0	0	0	0
+GO:0005991	0	0	8	0	0
+GO:0006002	4	2	0	0	0
+GO:0006007	0	0	0	0	2
+GO:0006011	0	0	9	0	0
+GO:0006012	0	0	2	0	0
+GO:0006071	2	0	0	0	2
+GO:0006090	4	0	4	0	0
+GO:0006091	0	4	0	0	2
+GO:0006096	2	8	4	0	0
+GO:0006097	4	0	0	0	0
+GO:0006099	0	0	2	0	0
+GO:0006106	0	0	0	0	4
+GO:0006207	4	0	0	0	0
+GO:0006221	6	0	0	0	0
+GO:0006259	0	4	0	0	0
+GO:0006260	0	8	2	0	10
+GO:0006265	10	0	0	2	4
+GO:0006268	0	0	4	0	0
+GO:0006281	7	8	18	0	6
+GO:0006298	4	0	0	0	2
+GO:0006302	0	0	4	0	0
+GO:0006310	4	1	6	2	0
+GO:0006313	14	6	1	10	0
+GO:0006351	4	0	0	0	0
+GO:0006352	0	4	4	0	0
+GO:0006355	11	8	4	15	10
+GO:0006364	4	0	0	0	2
+GO:0006383	0	0	0	0	2
+GO:0006396	8	0	4	0	2
+GO:0006400	0	0	0	0	6
+GO:0006412	16	2	7	10	8
+GO:0006414	0	0	0	2	0
+GO:0006415	0	4	0	0	0
+GO:0006418	4	8	4	0	8
+GO:0006419	4	0	0	0	0
+GO:0006420	1	0	0	0	0
+GO:0006421	2	0	0	0	0
+GO:0006427	0	0	0	4	0
+GO:0006429	12	0	0	0	0
+GO:0006431	4	0	0	0	0
+GO:0006432	2	0	0	0	0
+GO:0006433	2	0	0	0	0
+GO:0006435	6	4	4	0	0
+GO:0006457	4	2	4	0	0
+GO:0006464	0	0	0	3	0
+GO:0006465	4	0	0	0	0
+GO:0006520	8	0	0	0	4
+GO:0006522	0	0	4	0	0
+GO:0006526	4	0	0	0	0
+GO:0006527	0	0	0	4	0
+GO:0006537	0	2	4	0	0
+GO:0006541	2	0	0	0	0
+GO:0006564	2	0	0	0	0
+GO:0006576	0	0	0	0	4
+GO:0006596	0	4	0	0	0
+GO:0006614	0	0	3	0	0
+GO:0006629	0	4	0	0	3
+GO:0006633	0	0	0	0	6
+GO:0006655	0	0	2	0	0
+GO:0006696	2	0	0	0	0
+GO:0006725	0	3	4	0	2
+GO:0006777	4	0	0	0	0
+GO:0006779	0	0	4	0	0
+GO:0006782	2	0	0	0	0
+GO:0006783	4	2	4	0	0
+GO:0006790	0	0	0	0	4
+GO:0006799	0	0	2	0	0
+GO:0006807	0	4	4	4	0
+GO:0006810	0	4	4	12	6
+GO:0006811	0	4	2	0	0
+GO:0006813	2	8	4	0	2
+GO:0006814	0	2	0	0	0
+GO:0006821	0	0	0	0	4
+GO:0006855	0	0	6	0	0
+GO:0006865	0	2	4	0	6
+GO:0006935	0	3	0	0	0
+GO:0007049	0	0	0	0	4
+GO:0007155	6	2	3	4	0
+GO:0007156	0	4	0	0	0
+GO:0007165	0	6	0	0	4
+GO:0008152	4	2	18	4	8
+GO:0008295	5	0	0	0	0
+GO:0008299	2	0	0	0	0
+GO:0008360	2	0	4	0	2
+GO:0008610	4	0	0	0	0
+GO:0008616	4	0	0	0	0
+GO:0008643	0	7	5	12	0
+GO:0008654	4	0	0	0	0
+GO:0009056	0	4	0	0	0
+GO:0009058	6	4	6	0	2
+GO:0009059	4	0	0	2	0
+GO:0009060	4	0	0	0	0
+GO:0009063	0	0	4	0	0
+GO:0009072	4	0	0	0	2
+GO:0009073	3	0	0	4	0
+GO:0009086	0	0	0	4	0
+GO:0009089	0	2	4	0	4
+GO:0009098	4	0	0	0	0
+GO:0009103	2	0	6	0	2
+GO:0009116	1	0	3	0	0
+GO:0009166	2	0	0	0	0
+GO:0009228	9	0	3	0	0
+GO:0009234	4	0	0	0	4
+GO:0009236	4	0	0	0	0
+GO:0009244	0	4	0	0	0
+GO:0009245	0	2	2	0	0
+GO:0009252	0	0	4	0	0
+GO:0009253	0	0	0	0	4
+GO:0009264	2	0	0	0	0
+GO:0009271	0	0	1	0	0
+GO:0009306	0	2	4	0	1
+GO:0009401	0	4	5	3	0
+GO:0009404	0	0	2	0	0
+GO:0009405	0	0	7	0	4
+GO:0009435	0	2	0	0	0
+GO:0009446	0	0	0	4	0
+GO:0009972	0	0	4	0	0
+GO:0009987	0	2	6	10	4
+GO:0010033	0	2	0	0	0
+GO:0010038	2	0	0	0	0
+GO:0010124	4	0	0	0	0
+GO:0010133	0	0	0	0	2
+GO:0010468	0	0	0	0	2
+GO:0015031	2	2	2	4	0
+GO:0015074	0	0	4	0	0
+GO:0015628	0	0	0	0	2
+GO:0015803	0	0	4	0	0
+GO:0015833	4	0	0	8	0
+GO:0015889	0	0	1	0	0
+GO:0015891	0	0	4	0	0
+GO:0015909	0	2	0	0	0
+GO:0015937	4	0	0	0	0
+GO:0015976	0	4	2	0	0
+GO:0015986	0	0	2	0	2
+GO:0016051	0	4	0	0	0
+GO:0016114	0	0	0	0	2
+GO:0016539	0	0	0	0	4
+GO:0017003	4	0	0	0	0
+GO:0017004	0	0	2	0	0
+GO:0018339	4	0	0	0	0
+GO:0019285	0	0	4	0	0
+GO:0019310	0	4	0	4	0
+GO:0019354	2	0	0	0	0
+GO:0019439	4	0	2	0	0
+GO:0019544	0	0	0	0	2
+GO:0019569	0	0	0	2	0
+GO:0019700	0	0	6	0	0
+GO:0019752	0	0	0	4	0
+GO:0022904	4	0	0	0	0
+GO:0030001	0	4	0	0	0
+GO:0030091	0	2	0	0	0
+GO:0030436	0	3	0	0	0
+GO:0031119	4	0	0	0	0
+GO:0033539	0	0	0	2	0
+GO:0034641	0	0	10	6	4
+GO:0042128	2	0	0	0	0
+GO:0042254	4	0	0	0	0
+GO:0042426	0	0	0	0	4
+GO:0042773	1	0	0	0	0
+GO:0043039	0	0	4	0	0
+GO:0043213	0	0	0	3	0
+GO:0043571	2	0	0	0	0
+GO:0044179	0	0	1	0	0
+GO:0044205	6	0	0	0	0
+GO:0044208	2	0	0	0	0
+GO:0044209	0	0	0	0	5
+GO:0044237	4	0	20	3	5
+GO:0044238	0	0	2	0	0
+GO:0044249	0	0	2	0	0
+GO:0044260	6	0	2	0	0
+GO:0044281	0	0	0	2	0
+GO:0045454	0	2	0	0	2
+GO:0045892	0	0	2	4	0
+GO:0045900	0	0	0	0	4
+GO:0046416	0	0	0	0	1
+GO:0046854	1	1	0	0	0
+GO:0050794	0	0	0	0	2
+GO:0051191	0	0	0	2	0
+GO:0051301	2	0	2	3	2
+GO:0055070	0	0	2	0	0
+GO:0055085	19	4	22	0	29
+GO:0055086	0	0	1	0	0
+GO:0055114	0	0	7	0	4
+GO:0055129	2	0	0	0	0
+GO:0070084	4	0	0	2	0
+GO:0070475	0	4	0	0	0
+GO:0071470	0	3	0	0	0
+GO:0071704	0	0	0	2	0
+GO:0071705	0	0	4	0	0
+GO:0071840	1	0	0	0	0
+GO:0071973	0	0	2	0	4
+GO:0097588	0	2	0	0	0
+GO:1901135	0	0	0	0	2
+GO:1901564	0	0	5	0	0
+GO:1902600	0	0	4	2	2
+GO:1902604	0	0	0	0	2

woltka/tests/test_table.py

@@ -553,6 +553,12 @@ def test_collapse_table(self):
         for i in range(4):
             self.assertListEqual(obs[i], exp[i])
 
+        # BIOM table
+        table_ = Table(*map(np.array, table))
+        obs = collapse_table(table_, mapping)
+        exp = Table(*map(np.array, exp))
+        self.assertEqual(obs.descriptive_equality(exp), 'Tables appear equal')
+
         # some missing, some extra
         mapping = {'G1': ['H1'], 'G2': ['H2'], 'G3': ['H3'], 'G9': ['H9']}
         obs = collapse_table(table, mapping)

woltka/util.py

@@ -108,9 +108,6 @@ def intize(num):
 
     Notes
     -----
-    This function is for demo purpose. Use `initize_list` or `initize_dict` to
-    save function-calling overhead in data-intensive operations.
-
     Limited by Python floating point arithmetic, odds are that when ambiguous
     assignment is on, counts will be sums of fractions and they can be rounded
     either up or down when converted to integers, depending how these numbers
@@ -161,6 +158,11 @@ def intize_list(lst):
     See Also
     --------
     intize
+
+    Notes
+    -----
+    Use this function instead of `initize` on large lists to save some
+    function-calling overhead.
     """
     for i, element in enumerate(lst):
         near = round(element * 2) / 2
@@ -183,6 +185,11 @@ def intize_dict(dic, zero=False):
     See Also
     --------
     intize
+
+    Notes
+    -----
+    Use this function instead of `initize` on large dictionaries to save some
+    function-calling overhead.
     """
     todel = []
     add_todel = todel.append